Generate six distinct visual directions for the same subject. Vary composition, lighting and mood across all six. Keep every output low quality: these are for choosing a direction, not for use.\n\nSubject: [SUBJECT]\nIntended final use: [WHERE IT WILL APPEAR]
Create a single reference image of [SUBJECT] on a plain neutral background. Describe the subject in exhaustive detail in your own words first, then generate. I will edit this same image repeatedly, so prioritise a clean, well-lit, unambiguous rendering over an interesting one.
Using the attached image, make exactly one change: [CHANGE]. Everything else must stay identical, including pose, framing, lighting direction and background. Do not reinterpret the subject. If the change cannot be made without altering anything else, say so instead of generating.
Produce a flat vector icon of [SUBJECT] as SVG. Single colour on transparent, no gradients, no text, no outline strokes thinner than 2 units. Remember SVG output needs n=1 and will not come back from the edit endpoint.
Create a short looping animation showing [ACTION]. The loop must be seamless: the final frame has to match the first. Keep the motion to one element and hold everything else still.
Render [PRODUCT] three times on the same seamless background with identical framing. Vary only the lighting: soft diffused, hard directional from upper left, and rim lighting from behind. Same camera position in all three.
Create a poster for [EVENT]. The only text is: [EXACT TEXT]. Render it exactly as written, no additions, no decorative lettering that changes the characters. Set it in a single weight with generous spacing and leave clear margins.
Using the attached image only as a reference for palette and lighting, create a new image of [NEW SUBJECT]. Do not reproduce the composition, the subject or any recognisable element of the reference. Take the colour relationships and the quality of light, nothing else.
Generate [SUBJECT] twice: once at minimum complexity and once at maximum. Describe in one line what actually changed between them so I can decide which setting this subject needs.
Produce a labelled diagram of [PROCESS] as SVG. Boxes and arrows only, no illustration, no shading. Labels: [LIST OF LABELS]. Keep every label horizontal and legible at 400 pixels wide.
Create [SUBJECT] as three separate images sized for a square post, a vertical story and a wide header. Compose each one for its own aspect ratio rather than cropping a single image. Keep the subject at the same scale relative to the frame.
Review the attached generated image against this list and tell me which items fail: text is spelled correctly, hands and extremities are anatomically plausible, no duplicated or floating elements, lighting direction is consistent across the whole frame, edges are clean against the background. List only the failures.